Journal
PRECISION AGRICULTURE
Volume 17, Issue 4, Pages 488-505Publisher
SPRINGER
DOI: 10.1007/s11119-016-9432-2
Keywords
Grapevine leafroll disease; Remote sensing; Spectral reflectance; Minimum noise fraction; Pixel purity index; Ant colony clustering algorithm
Categories
Funding
- Key Project of Scientific Research Institutions of Higher Learning in Ningxia [NGY2013005]
Ask authors/readers for more resources
Grapevine leafroll disease (GLD) is a virus disease that quickly propagates through vineyards under appropriate weather conditions and can reduce grape production worldwide by nearly 60 %. Therefore, the accurate diagnosis and reliable evaluation of GLD distribution, particularly at the early stage of GLD infection, is important to prevent the spread of this disease. This study applied the ant colony clustering algorithm (ACCA) to detect GLD spectral anomalies on 4 GLD-infected vineyards according to multi-spectral imagery for precision disease management. GLD was classified into three stages: GLD1, GLD2 and GLD3 according to its infection severity. An 11-index feature vector and its stacked image were generated to enhance the spectral differences and spectral discrimination between diseased and healthy grapevines. ACCA was then designed to solve the fuzziness of the multi-spectral image for GLD-infected grapevines and successfully identify GLD from healthy grapevines. Finally, a field survey with 49 samples and pixel purity index technology were applied to validate the effectiveness, efficiency and accuracy of ACCA. Field results indicated that an early stage of the GLD infection (GLD1) could be successfully discriminated from GLD2-, GLD3- and non-infected grapevines. The classification accuracies of non-, GLD1-, GLD2- and GLD3-infected grapevines were 94.4, 75, 84.6 and 83.3 %, respectively. Hence, the method based on an 11-index image and ACCA may significantly detect GLD at an early stage from healthy grapevines for precision disease management at the field level.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available